Figure 2
From: Deepfake smiles matter less—the psychological and neural impact of presumed AI-generated faces
Facial expression ratings (a) and response times (b) depending on Information and Emotion conditions. Violin plots are based on by-participant means per condition. Black points show condition means and error bars represent 95% confidence intervals with inter-subject variance removed. Correspondence of rating values in (a): 1 = very negative, 51 = neutral, 101 = very positive. Asterisks indicate significant differences between the respective conditions and a significant interaction between Information and Positive Emotion (labeled as “IA” in the plot) observed in the analyses: *** p < .001, ** p < .01, * p < .05. An alternative illustration showing by-participant differences between Real and Fake conditions per Emotion is provided in Supplement Fig. S5.
